A　simplified　adaptive　particle　filter　was　proposed　based　on　the　K-L　sampling　to　overcome　the　drawback　of　computational　burden　in　the　traditional　adaptive　particle　filter.　The　new　algorithm　calculates　k　value　by　separating　it　from　the　sampling　process.　Therefore,　the　complex　and　computational　burden　of　the　algorithm　are　efficiently　reduced,　and　the　infinite　circulation　is　avoided.　The　simplified　adaptive　particle　filter　was　applied　to　the　SINS　non-linear　alignment　with　large　azimuth.　Compared　with　Extended　Kalman　Filter,　standard　particle　filter　and　traditional　adaptive　particle　filter,　the　simulated　data　show　that　simplified　adaptive　particle　filter　results　in　more　efficient　alignment　without　reducing　the　accuracy　of　the　SINS　alignment.　Then　the　simplified　adaptive　particle　filter　is　more　suitable　to　the　SINS　non-linear　alignment　in　dynamic　situation.